Occurrence data

Explore occurrence records

occ.sterechinus <- read.csv("data/occurrences_sterechinus.csv", header=T, sep=";")
head(occ.sterechinus)
##   decimalLatitude decimalLongitude
## 1       -77.85000         166.6667
## 2       -77.83333         166.5667
## 3       -77.83330         166.5000
## 4       -77.83330         166.5500
## 5       -77.67000         -35.5000
## 6       -77.66667         -35.5000
ggplot(occ.sterechinus, aes(x=decimalLongitude, y=decimalLatitude)) +
  geom_point(stat="identity", colour="blue", alpha=0.5)

Environmental layers

depth <- raster("data/environmental_layers/depth.nc")
geomorphology <- raster("data/environmental_layers/geomorphology.nc")
ice_cover_min <- raster("data/environmental_layers/ice_cover_min.nc")
ice_cover_max <- raster("data/environmental_layers/ice_cover_max.nc")
ice_thickness_min <- raster("data/environmental_layers/ice_thickness_min.nc")
ice_thickness_max <- raster("data/environmental_layers/ice_thickness_max.nc")
mixed_layer_depth <- raster("data/environmental_layers/mixed_layer_depth.nc")
POC_2005_2012_min <- raster("data/environmental_layers/POC_2005_2012_min.nc")
POC_2005_2012_max <- raster("data/environmental_layers/POC_2005_2012_max.nc")
roughness <- raster("data/environmental_layers/roughness.nc")
sediments <- raster("data/environmental_layers/sediments.nc")
seafloor_current_speed <- raster("data/environmental_layers/seafloor_current_speed.nc")
seafloor_sali_2005_2012_min <- raster("data/environmental_layers/seafloor_sali_2005_2012_min.nc")
seafloor_sali_2005_2012_max <- raster("data/environmental_layers/seafloor_sali_2005_2012_max.nc")
seafloor_temp_2005_2012_min <- raster("data/environmental_layers/seafloor_temp_2005_2012_min.nc")
seafloor_temp_2005_2012_max <- raster("data/environmental_layers/seafloor_temp_2005_2012_max.nc")
slope <- raster("data/environmental_layers/slope.nc")

predictors_stack <- stack(depth, geomorphology,ice_cover_min, ice_cover_max, ice_thickness_min, ice_thickness_max, mixed_layer_depth, POC_2005_2012_min, POC_2005_2012_max, roughness, sediments, seafloor_current_speed, seafloor_sali_2005_2012_max, seafloor_temp_2005_2012_max, slope)

Pixel resolution of environmental layers

Convert pixel resolution of predictors_stack with raster::aggregate. Choices: 0.1, 1, 10?

# check current pixel resolution of predictors_stack
predictors_stack_5 <- aggregate(predictors_stack, fact=5)
predictors_stack_10 <- aggregate(predictors_stack, fact=10)
plot(predictors_stack)

plot(predictors_stack_5)

plot(predictors_stack_10)

# pass predictors_stack to run() function
source("scripts/run_yOur_SDM.R")
res_1 <- run(predictors_stack, "res_0.1")
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1240 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.9827 
## tolerance is fixed at  0.001 
## ntrees resid. dev. 
## 50    0.687 
## now adding trees... 
## 100   0.5771 
## 150   0.5199 
## 200   0.4909 
## 250   0.4768 
## 300   0.468 
## 350   0.4628 
## 400   0.4592 
## 450   0.4563 
## 500   0.4565 
## 550   0.4565 
## 600   0.4557 
## 650   0.4557 
## 700   0.4568 
## 750   0.4581 
## 800   0.4606 
## 850   0.4614 
## 900   0.4632 
## 950   0.4645 
## 1000   0.4668 
## 1050   0.4692 
## 1100   0.4719 
## 1150   0.4737
## fitting final gbm model with a fixed number of 650 trees for id

## 
## mean total deviance = 0.983 
## mean residual deviance = 0.313 
##  
## estimated cv deviance = 0.456 ; se = 0.064 
##  
## training data correlation = 0.851 
## cv correlation =  0.717 ; se = 0.029 
##  
## training data AUC score = 0.983 
## cv AUC score = 0.949 ; se = 0.009 
##  
## elapsed time -  0.1 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1240 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.9827 
## tolerance is fixed at  0.001 
## ntrees resid. dev. 
## 50    0.7 
## now adding trees... 
## 100   0.5955 
## 150   0.5426 
## 200   0.5137 
## 250   0.4993 
## 300   0.4913 
## 350   0.4852 
## 400   0.4806 
## 450   0.4781 
## 500   0.476 
## 550   0.4736 
## 600   0.4718 
## 650   0.4709 
## 700   0.4708 
## 750   0.472 
## 800   0.4716 
## 850   0.4722 
## 900   0.4733 
## 950   0.4737 
## 1000   0.4751 
## 1050   0.4767 
## 1100   0.478 
## 1150   0.479 
## 1200   0.4812 
## 1250   0.4817
## fitting final gbm model with a fixed number of 700 trees for id

## 
## mean total deviance = 0.983 
## mean residual deviance = 0.292 
##  
## estimated cv deviance = 0.471 ; se = 0.062 
##  
## training data correlation = 0.866 
## cv correlation =  0.71 ; se = 0.024 
##  
## training data AUC score = 0.987 
## cv AUC score = 0.945 ; se = 0.01 
##  
## elapsed time -  0.1 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1240 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.9827 
## tolerance is fixed at  0.001 
## ntrees resid. dev. 
## 50    0.6971 
## now adding trees... 
## 100   0.5885 
## 150   0.5365 
## 200   0.5097 
## 250   0.493 
## 300   0.4839 
## 350   0.4784 
## 400   0.4755 
## 450   0.4735 
## 500   0.4727 
## 550   0.4713 
## 600   0.4714 
## 650   0.4721 
## 700   0.472 
## 750   0.4728 
## 800   0.4742 
## 850   0.4749 
## 900   0.4761 
## 950   0.4773 
## 1000   0.4795 
## 1050   0.4809 
## 1100   0.4826 
## 1150   0.4833 
## 1200   0.4855
## fitting final gbm model with a fixed number of 550 trees for id

## 
## mean total deviance = 0.983 
## mean residual deviance = 0.32 
##  
## estimated cv deviance = 0.471 ; se = 0.035 
##  
## training data correlation = 0.848 
## cv correlation =  0.709 ; se = 0.014 
##  
## training data AUC score = 0.982 
## cv AUC score = 0.944 ; se = 0.006 
##  
## elapsed time -  0.09 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1240 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.9827 
## tolerance is fixed at  0.001 
## ntrees resid. dev. 
## 50    0.7376 
## now adding trees... 
## 100   0.6343 
## 150   0.5842 
## 200   0.5577 
## 250   0.5444 
## 300   0.5364 
## 350   0.534 
## 400   0.5324 
## 450   0.5321 
## 500   0.5319 
## 550   0.5322 
## 600   0.5325 
## 650   0.533 
## 700   0.5347 
## 750   0.5361 
## 800   0.5374 
## 850   0.5385 
## 900   0.539 
## 950   0.5407 
## 1000   0.5436 
## 1050   0.5456 
## 1100   0.5478 
## 1150   0.5496
## fitting final gbm model with a fixed number of 500 trees for id

## 
## mean total deviance = 0.983 
## mean residual deviance = 0.357 
##  
## estimated cv deviance = 0.532 ; se = 0.033 
##  
## training data correlation = 0.825 
## cv correlation =  0.667 ; se = 0.026 
##  
## training data AUC score = 0.977 
## cv AUC score = 0.929 ; se = 0.007 
##  
## elapsed time -  0.09 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1240 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.9827 
## tolerance is fixed at  0.001 
## ntrees resid. dev. 
## 50    0.7683 
## now adding trees... 
## 100   0.6434 
## 150   0.584 
## 200   0.5565 
## 250   0.5454 
## 300   0.5393 
## 350   0.5354 
## 400   0.5319 
## 450   0.5301 
## 500   0.5279 
## 550   0.5261 
## 600   0.5259 
## 650   0.5268 
## 700   0.5271 
## 750   0.5303 
## 800   0.5312 
## 850   0.5342 
## 900   0.5359 
## 950   0.5372 
## 1000   0.5381 
## 1050   0.539 
## 1100   0.5411 
## 1150   0.5435
## fitting final gbm model with a fixed number of 600 trees for id

## 
## mean total deviance = 0.983 
## mean residual deviance = 0.324 
##  
## estimated cv deviance = 0.526 ; se = 0.024 
##  
## training data correlation = 0.844 
## cv correlation =  0.724 ; se = 0.021 
##  
## training data AUC score = 0.982 
## cv AUC score = 0.942 ; se = 0.004 
##  
## elapsed time -  0.09 minutes

res_5 <- run(predictors_stack_5, "res_0.5")
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1155 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.7886 
## tolerance is fixed at  8e-04 
## ntrees resid. dev. 
## 50    0.5954 
## now adding trees... 
## 100   0.5271 
## 150   0.5002 
## 200   0.4895 
## 250   0.4851 
## 300   0.4825 
## 350   0.4819 
## 400   0.4836 
## 450   0.4866 
## 500   0.4891 
## 550   0.4923 
## 600   0.496 
## 650   0.4991 
## 700   0.5028 
## 750   0.5074 
## 800   0.5122 
## 850   0.5167 
## 900   0.5223 
## 950   0.5256 
## 1000   0.53
## fitting final gbm model with a fixed number of 350 trees for id

## 
## mean total deviance = 0.789 
## mean residual deviance = 0.358 
##  
## estimated cv deviance = 0.482 ; se = 0.037 
##  
## training data correlation = 0.744 
## cv correlation =  0.536 ; se = 0.058 
##  
## training data AUC score = 0.965 
## cv AUC score = 0.914 ; se = 0.012 
##  
## elapsed time -  0.09 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1155 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.7886 
## tolerance is fixed at  8e-04 
## ntrees resid. dev. 
## 50    0.5805 
## now adding trees... 
## 100   0.5099 
## 150   0.4805 
## 200   0.4652 
## 250   0.4568 
## 300   0.4515 
## 350   0.448 
## 400   0.4461 
## 450   0.4445 
## 500   0.4442 
## 550   0.4451 
## 600   0.4452 
## 650   0.4454 
## 700   0.4471 
## 750   0.4482 
## 800   0.4488 
## 850   0.4496 
## 900   0.4511 
## 950   0.4526 
## 1000   0.4531 
## 1050   0.4549 
## 1100   0.4561 
## 1150   0.4588
## fitting final gbm model with a fixed number of 500 trees for id

## 
## mean total deviance = 0.789 
## mean residual deviance = 0.307 
##  
## estimated cv deviance = 0.444 ; se = 0.067 
##  
## training data correlation = 0.795 
## cv correlation =  0.631 ; se = 0.05 
##  
## training data AUC score = 0.976 
## cv AUC score = 0.935 ; se = 0.014 
##  
## elapsed time -  0.1 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1155 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.7886 
## tolerance is fixed at  8e-04 
## ntrees resid. dev. 
## 50    0.5923 
## now adding trees... 
## 100   0.5254 
## 150   0.496 
## 200   0.4825 
## 250   0.4758 
## 300   0.4718 
## 350   0.4699 
## 400   0.469 
## 450   0.4686 
## 500   0.4685 
## 550   0.4704 
## 600   0.4721 
## 650   0.4725 
## 700   0.4744 
## 750   0.4778 
## 800   0.4805 
## 850   0.482 
## 900   0.4839 
## 950   0.4847 
## 1000   0.4871 
## 1050   0.4892
## fitting final gbm model with a fixed number of 500 trees for id

## 
## mean total deviance = 0.789 
## mean residual deviance = 0.308 
##  
## estimated cv deviance = 0.468 ; se = 0.021 
##  
## training data correlation = 0.798 
## cv correlation =  0.57 ; se = 0.039 
##  
## training data AUC score = 0.977 
## cv AUC score = 0.915 ; se = 0.012 
##  
## elapsed time -  0.09 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1155 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.7886 
## tolerance is fixed at  8e-04 
## ntrees resid. dev. 
## 50    0.6213 
## now adding trees... 
## 100   0.5581 
## 150   0.533 
## 200   0.5231 
## 250   0.5197 
## 300   0.5177 
## 350   0.5164 
## 400   0.5173 
## 450   0.5173 
## 500   0.5175 
## 550   0.518 
## 600   0.5203 
## 650   0.5219 
## 700   0.5239 
## 750   0.5261 
## 800   0.5286 
## 850   0.5314 
## 900   0.5337 
## 950   0.5369 
## 1000   0.5395 
## 1050   0.5421
## fitting final gbm model with a fixed number of 350 trees for id

## 
## mean total deviance = 0.789 
## mean residual deviance = 0.371 
##  
## estimated cv deviance = 0.516 ; se = 0.028 
##  
## training data correlation = 0.739 
## cv correlation =  0.53 ; se = 0.058 
##  
## training data AUC score = 0.964 
## cv AUC score = 0.899 ; se = 0.016 
##  
## elapsed time -  0.09 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1155 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.7886 
## tolerance is fixed at  8e-04 
## ntrees resid. dev. 
## 50    0.662 
## now adding trees... 
## 100   0.5906 
## 150   0.562 
## 200   0.551 
## 250   0.5496 
## 300   0.5479 
## 350   0.5476 
## 400   0.5506 
## 450   0.5516 
## 500   0.5555 
## 550   0.5576 
## 600   0.5596 
## 650   0.5625 
## 700   0.5646 
## 750   0.5688 
## 800   0.5731 
## 850   0.5759 
## 900   0.58 
## 950   0.5833 
## 1000   0.5864
## fitting final gbm model with a fixed number of 350 trees for id

## 
## mean total deviance = 0.789 
## mean residual deviance = 0.373 
##  
## estimated cv deviance = 0.548 ; se = 0.075 
##  
## training data correlation = 0.733 
## cv correlation =  0.537 ; se = 0.053 
##  
## training data AUC score = 0.963 
## cv AUC score = 0.894 ; se = 0.019 
##  
## elapsed time -  0.08 minutes

res_10 <- run(predictors_stack_10, "res_1")
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1110 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.6462 
## tolerance is fixed at  6e-04 
## ntrees resid. dev. 
## 50    0.5044 
## now adding trees... 
## 100   0.4554 
## 150   0.4339 
## 200   0.4263 
## 250   0.4241 
## 300   0.4224 
## 350   0.4228 
## 400   0.4229 
## 450   0.4235 
## 500   0.4252 
## 550   0.427 
## 600   0.4286 
## 650   0.4315 
## 700   0.4333 
## 750   0.4352 
## 800   0.4375 
## 850   0.4402 
## 900   0.4431 
## 950   0.4465 
## 1000   0.4486
## fitting final gbm model with a fixed number of 300 trees for id

## 
## mean total deviance = 0.646 
## mean residual deviance = 0.337 
##  
## estimated cv deviance = 0.422 ; se = 0.043 
##  
## training data correlation = 0.675 
## cv correlation =  0.518 ; se = 0.046 
##  
## training data AUC score = 0.959 
## cv AUC score = 0.916 ; se = 0.021 
##  
## elapsed time -  0.08 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1110 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.6462 
## tolerance is fixed at  6e-04 
## ntrees resid. dev. 
## 50    0.5139 
## now adding trees... 
## 100   0.475 
## 150   0.4608 
## 200   0.4557 
## 250   0.4561 
## 300   0.4581 
## 350   0.4617 
## 400   0.4649 
## 450   0.4704 
## 500   0.4739 
## 550   0.4784 
## 600   0.4828 
## 650   0.4879 
## 700   0.4936 
## 750   0.4981 
## 800   0.5028 
## 850   0.5079 
## 900   0.5125 
## 950   0.5171 
## 1000   0.522
## fitting final gbm model with a fixed number of 200 trees for id

## 
## mean total deviance = 0.646 
## mean residual deviance = 0.364 
##  
## estimated cv deviance = 0.456 ; se = 0.04 
##  
## training data correlation = 0.656 
## cv correlation =  0.442 ; se = 0.034 
##  
## training data AUC score = 0.951 
## cv AUC score = 0.903 ; se = 0.011 
##  
## elapsed time -  0.07 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1110 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.6462 
## tolerance is fixed at  6e-04 
## ntrees resid. dev. 
## 50    0.5054 
## now adding trees... 
## 100   0.466 
## 150   0.449 
## 200   0.442 
## 250   0.4421 
## 300   0.4429 
## 350   0.445 
## 400   0.4479 
## 450   0.4488 
## 500   0.4515 
## 550   0.4551 
## 600   0.4597 
## 650   0.4626 
## 700   0.4673 
## 750   0.471 
## 800   0.4746 
## 850   0.4796 
## 900   0.4841 
## 950   0.4881 
## 1000   0.4914
## fitting final gbm model with a fixed number of 200 trees for id

## 
## mean total deviance = 0.646 
## mean residual deviance = 0.355 
##  
## estimated cv deviance = 0.442 ; se = 0.03 
##  
## training data correlation = 0.672 
## cv correlation =  0.446 ; se = 0.028 
##  
## training data AUC score = 0.955 
## cv AUC score = 0.897 ; se = 0.009 
##  
## elapsed time -  0.07 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1110 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.6462 
## tolerance is fixed at  6e-04 
## ntrees resid. dev. 
## 50    0.517 
## now adding trees... 
## 100   0.4747 
## 150   0.456 
## 200   0.4489 
## 250   0.4461 
## 300   0.4451 
## 350   0.4444 
## 400   0.4447 
## 450   0.4459 
## 500   0.4474 
## 550   0.4488 
## 600   0.4498 
## 650   0.4518 
## 700   0.4545 
## 750   0.4566 
## 800   0.4596 
## 850   0.4614 
## 900   0.4632 
## 950   0.4658 
## 1000   0.4686
## fitting final gbm model with a fixed number of 350 trees for id

## 
## mean total deviance = 0.646 
## mean residual deviance = 0.333 
##  
## estimated cv deviance = 0.444 ; se = 0.028 
##  
## training data correlation = 0.69 
## cv correlation =  0.453 ; se = 0.051 
##  
## training data AUC score = 0.958 
## cv AUC score = 0.895 ; se = 0.013 
##  
## elapsed time -  0.07 minutes 
## 
##  
##  GBM STEP - version 2.9 
##  
## Performing cross-validation optimisation of a boosted regression tree model 
## for id and using a family of bernoulli 
## Using 1110 observations and 15 predictors 
## loading user-supplied fold vector 
## creating 4 initial models of 50 trees 
## 
##  folds are stratified by prevalence 
## total mean deviance =  0.6462 
## tolerance is fixed at  6e-04 
## ntrees resid. dev. 
## 50    0.5426 
## now adding trees... 
## 100   0.5018 
## 150   0.4887 
## 200   0.4854 
## 250   0.4858 
## 300   0.4857 
## 350   0.49 
## 400   0.4949 
## 450   0.5009 
## 500   0.5077 
## 550   0.5127 
## 600   0.5175 
## 650   0.5217 
## 700   0.5273 
## 750   0.533 
## 800   0.5386 
## 850   0.5446 
## 900   0.5504 
## 950   0.5565 
## 1000   0.5628
## fitting final gbm model with a fixed number of 200 trees for id

## 
## mean total deviance = 0.646 
## mean residual deviance = 0.369 
##  
## estimated cv deviance = 0.485 ; se = 0.062 
##  
## training data correlation = 0.639 
## cv correlation =  0.416 ; se = 0.048 
##  
## training data AUC score = 0.949 
## cv AUC score = 0.888 ; se = 0.013 
##  
## elapsed time -  0.07 minutes

res_1
## [[1]]
##          depth geomorphology ice_cover_min ice_cover_max ice_thickness_min
## CtM  17.930666     1.2366647     0.1303194     20.346240        0.05250484
## CtSD  6.307375     0.1522527     0.1194673      4.232842        0.09175831
##      ice_thickness_max mixed_layer_depth POC_2005_2012_min
## CtM          10.203085          2.572348         5.8195153
## CtSD          2.444769          1.351595         0.9420132
##      POC_2005_2012_max roughness sediments seafloor_current_speed
## CtM          24.750621  4.420418 2.7557811              3.6960056
## CtSD          7.205073  0.754408 0.2060642              0.8038036
##      seafloor_sali_2005_2012_max seafloor_temp_2005_2012_max    slope
## CtM                    0.8421448                   2.7325400 2.511146
## CtSD                   0.3432618                   0.9939546 1.023322
## 
## [[2]]
##                       [,1]       [,2]       [,3]       [,4]       [,5]
## AUC              0.9558000  0.9330000  0.9361000  0.9722000  0.9305000
## COR              0.6992987  0.6636518  0.7071519  0.7978194  0.6885880
## TSS              0.6833179  0.5692363  0.5963489  0.7602871  0.5137931
## maxSSS           0.2857768  0.4591752  0.6050939  0.3014095  0.4262716
## valid_test_data 88.4146341 86.6421569 87.7862595 90.6976744 84.9315068
## prop_test        9.5833333 70.4166667 12.0833333  7.9166667 12.5000000
##                       [,6]       [,7]       [,8]       [,9]      [,10]
## AUC              0.9713000  0.9498000  0.9273000  0.9305000  0.9453000
## COR              0.7442441  0.7540633  0.6528057  0.6812191  0.7282217
## TSS              0.8524668  0.5913978  0.5729266  0.6677165  0.5612270
## maxSSS           0.4044831  0.6222833  0.3543146  0.4056526  0.4633763
## valid_test_data 91.4893617 87.1794872 86.7224880 89.5966030 86.4321608
## prop_test        7.0833333 10.0000000 70.4166667 37.5000000 37.9166667
##                      [,11]      [,12]      [,13]      [,14]      [,15]
## AUC              0.9427000  0.9592000  0.9362000  0.9396000  0.9085000
## COR              0.6898372  0.7377930  0.6842144  0.7190538  0.5963976
## TSS              0.6148148  0.7703037  0.6436905  0.5942961  0.4773810
## maxSSS           0.5319660  0.3623791  0.3334083  0.4928258  0.4408957
## valid_test_data 87.6543210 91.3875598 88.4297521 87.5930521 83.8028169
## prop_test       11.2500000 13.3333333 35.0000000 38.3333333 12.5000000
##                      [,16]      [,17]      [,18]      [,19]      [,20]
## AUC              0.9335000  0.9407000  0.9524000  0.9374000  0.9379000
## COR              0.6688055  0.7049988  0.7851446  0.6896559  0.7171650
## TSS              0.6799601  0.6382372  0.5873016  0.6559345  0.6397569
## maxSSS           0.3999113  0.4908271  0.5040245  0.5137912  0.4746118
## valid_test_data 89.0995261 88.7254902 86.6666667 89.0394089 89.0243902
## prop_test       14.1666667 15.8333333  7.5000000 61.6666667 15.0000000
## 
## [[3]]
## class      : RasterLayer 
## dimensions : 350, 3600, 1260000  (nrow, ncol, ncell)
## resolution : 0.1, 0.1  (x, y)
## extent     : -180, 180, -80, -45  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0 
## source     : memory
## names      : layer 
## values     : 0.008006955, 0.9354443  (min, max)
res_5
## [[1]]
##          depth geomorphology ice_cover_min ice_cover_max ice_thickness_min
## CtM  18.325566      3.453612     0.6106815     20.225266        0.06408695
## CtSD  6.225785      1.678995     0.7345645      4.123862        0.13361150
##      ice_thickness_max mixed_layer_depth POC_2005_2012_min
## CtM           8.704239          2.254451         12.617786
## CtSD          3.688243          1.091642          3.486626
##      POC_2005_2012_max roughness sediments seafloor_current_speed
## CtM          12.374563  4.348182 1.5513819                3.73427
## CtSD          3.471007  1.072554 0.9120895                1.34220
##      seafloor_sali_2005_2012_max seafloor_temp_2005_2012_max    slope
## CtM                     2.601947                    6.004284 3.129683
## CtSD                    1.022258                    1.398749 1.079657
## 
## [[2]]
##                       [,1]       [,2]       [,3]       [,4]       [,5]
## AUC              0.9147000  0.8910000  0.9465000  0.9019000  0.9507000
## COR              0.4195777  0.4951637  0.6939180  0.5336381  0.7064927
## TSS              0.5277934  0.3744476  0.4687637  0.4207921  0.6971596
## maxSSS           0.2407876  0.3214172  0.4952894  0.3874982  0.3901202
## valid_test_data 88.6666667 85.3562005 87.6923077 86.3247863 93.1034483
## prop_test        8.3870968 67.7419355 13.5483871 10.3225806 12.2580645
##                       [,6]       [,7]       [,8]       [,9]      [,10]
## AUC              0.9619000  0.9277000  0.8998000  0.9329000  0.9339000
## COR              0.6865479  0.6425200  0.4877488  0.5997191  0.6495390
## TSS              0.7467532  0.4641944  0.4361249  0.5259248  0.4834078
## maxSSS           0.2323910  0.4399640  0.3115352  0.2924519  0.4462696
## valid_test_data 91.8699187 87.1559633 86.8894602 89.0214797 87.9795396
## prop_test        7.0967742 10.9677419 69.6774194 36.1290323 39.3548387
##                      [,11]      [,12]      [,13]      [,14]      [,15]
## AUC              0.8832000  0.9083000  0.8998000  0.9148000  0.9281000
## COR              0.4640875  0.5647255  0.5133615  0.6379993  0.5943235
## TSS              0.4775414  0.4656627  0.4470613  0.4840809  0.3903509
## maxSSS           0.4087196  0.3088708  0.4539722  0.3527552  0.3412696
## valid_test_data 88.0503145 87.6344086 87.2685185 87.8787879 85.6115108
## prop_test       11.6129032 12.9032258 31.6129032 41.9354839 12.2580645
##                      [,16]      [,17]      [,18]      [,19]      [,20]
## AUC              0.8536000  0.8943000  0.8602000  0.8749000  0.9486000
## COR              0.3723716  0.4479509  0.5439102  0.4720726  0.6831350
## TSS              0.2612267  0.4030303  0.1987578  0.4118551  0.5507246
## maxSSS           0.2923341  0.3693720  0.4358779  0.2610224  0.4163966
## valid_test_data 83.5106383 86.7403315 76.6666667 86.3213811 89.4409938
## prop_test       14.1935484 10.3225806  9.0322581 65.8064516 14.8387097
## 
## [[3]]
## class      : RasterLayer 
## dimensions : 70, 720, 50400  (nrow, ncol, ncell)
## resolution : 0.5, 0.5  (x, y)
## extent     : -180, 180, -80, -45  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0 
## source     : memory
## names      : layer 
## values     : 0.01162865, 0.8395866  (min, max)
res_10
## [[1]]
##          depth geomorphology ice_cover_min ice_cover_max ice_thickness_min
## CtM  10.941070      3.713193    0.07883493     25.475356                 0
## CtSD  1.494837      1.387697    0.11893146      4.468935                 0
##      ice_thickness_max mixed_layer_depth POC_2005_2012_min
## CtM           3.160188          1.863450         17.612380
## CtSD          1.111991          1.375602          4.086493
##      POC_2005_2012_max roughness sediments seafloor_current_speed
## CtM          12.137185   4.46075 1.3268326              3.1796923
## CtSD          5.388202   2.61066 0.3380201              0.4709964
##      seafloor_sali_2005_2012_max seafloor_temp_2005_2012_max     slope
## CtM                     3.349486                    9.487713 3.2138700
## CtSD                    1.645406                    3.023571 0.9886124
## 
## [[2]]
##                       [,1]       [,2]       [,3]       [,4]       [,5]
## AUC              0.9386000  0.8529000  0.9253000  0.9457000  0.9046000
## COR              0.4557411  0.4215006  0.5922593  0.6030513  0.4283150
## TSS              0.6065163  0.3460458  0.3586674  0.4203013  0.2641026
## maxSSS           0.2785036  0.3585402  0.3317644  0.3075857  0.2542343
## valid_test_data 92.2535211 88.5989011 87.6923077 90.0000000 86.8965517
## prop_test        8.1818182 64.5454545 15.4545455 11.8181818 13.6363636
##                       [,6]       [,7]       [,8]       [,9]      [,10]
## AUC              0.9229000  0.9121000  0.8712000  0.9145000  0.9027000
## COR              0.4827105  0.5038063  0.3521728  0.5276575  0.4322445
## TSS              0.4345794  0.3996004  0.3014568  0.4305223  0.2834986
## maxSSS           0.3366517  0.2759588  0.2693344  0.3157113  0.2388737
## valid_test_data 90.4347826 89.2156863 87.5668449 89.7959184 86.5979381
## prop_test        7.2727273 10.0000000 69.0909091 35.4545455 41.8181818
##                      [,11]      [,12]      [,13]      [,14]      [,15]
## AUC              0.9000000  0.8712000  0.8778000  0.8944000  0.8737000
## COR              0.4026478  0.4220008  0.4342483  0.4170871  0.3607531
## TSS              0.3408046  0.3927885  0.2574526  0.2392720  0.2842713
## maxSSS           0.2509619  0.3140128  0.3291845  0.2922217  0.3315728
## valid_test_data 88.5350318 89.5953757 87.1604938 85.7506361 87.5912409
## prop_test       10.9090909 11.8181818 32.7272727 40.9090909 10.0000000
##                      [,16]      [,17]      [,18]      [,19]      [,20]
## AUC              0.9324000  0.9250000  0.8878000  0.8698000  0.8704000
## COR              0.5983217  0.4258726  0.5478889  0.3401614  0.3484633
## TSS              0.4490446  0.5500000  0.1437908  0.1730864  0.1997636
## maxSSS           0.3057400  0.3435409  0.3888977  0.2649450  0.3153624
## valid_test_data 90.2857143 91.0714286 81.6666667 85.0622407 84.9056604
## prop_test       16.3636364  7.2727273  8.1818182 68.1818182 16.3636364
## 
## [[3]]
## class      : RasterLayer 
## dimensions : 35, 360, 12600  (nrow, ncol, ncell)
## resolution : 1, 1  (x, y)
## extent     : -180, 180, -80, -45  (xmin, xmax, ymin, ymax)
## crs        : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0 
## source     : memory
## names      : layer 
## values     : 0.01888601, 0.6893824  (min, max)